Goals of the Internet
Reaction to David D. Clark’s “The Design Philosophy of the DARPA Internet Protocols“
The Internet has exponentially grown from its inception up to now, roughly four decades later, with little modification. A time-tested design, it has been the vehicle of varying information, no matter the kind of transmission medium the pathway is. The TCP/IP suite would not have developed and remained to be the standard until now if not for the designer’s objectives of following an ordered set of goals.
First and foremost, the goal was to interconnect the then-existing packet-switched ARPA and ARPANET networks. This determined the structure of the Internet: a packet-switched communications facility connected using gateways. The next goal was survivability — two communicating hosts should be able to converse despite network disruption, as long as there is any path between them. The Internet, after all, is a communications facility funded by the DARPA and should serve its purpose even during wartime. This required that state information be stored at the end hosts in what is called “fate-sharing” model — state information, if stored in the gateways, is useless anyway if any of the communicating hosts lost its connection (see end-to-end argument). The designers also realized that different services are required, which cannot be supported by a single protocol, so TCP/IP was split into two layers. The rest of the goals were given less priority but achieved to a certain degree: distributed management, cost-effectiveness, easy host attachment, and accountability.
The widespread use of the Internet past its military origins is currently challenging its design limits. As noted in the paper, a different protocol would have resulted had the priorities been different. The needs at the time of TCP/IP’s inception are quite different from the current uses demanded of it. Today, there are so many applications for the Internet, from web browsing, e-commerce, multimedia streaming, etc. and each of these applications have different requirements in terms of service. Aside from the applications, the transmission media are also changing, with wired and wireless connections available. This means more possible environments and conditions on which these facilities can be deployed. The fast pace of technology has also given birth to many other devices that can connect to the Internet. All these will have to be accommodated and so the Internet, in its evolution, will have to remain flexible and extensible to support future developments.
The latter goals which did not receive as much attention as the first three goals are now requiring attention, especially now that the use of the Internet has been passed down from the military to commercial and entertainment purposes. Security is now a big issue — back then, the pioneers had the same objectives and the connected nodes can be trusted; the same can no longer be said today, with all the cyberattacks and malicious users in the Internet. Accountability is also more important now, as businesses and consumers would like to know how resources are being used and whether they are used fairly.
Still, I believe that the current Internet design need not be scrapped or overhauled from ground up; network research can focus on the other individual layers or study cross-layer approaches.
Filed under: Computer Networks, Internet | Leave a Comment
No Responses Yet to “Goals of the Internet”